Interface cérebro computador por potencial evocado visual de regime permanente baseada em coerência múltipla e teste F normalizado

Abstract

Brain Computer Interfaces (BCI) are systems that are capable of translating information from the neuronal activity of a subject into controlling signals for external devices to be used for communication and motor impairment rehabilitation. The active interface developed here uses evoked potentials of periodic sensory stimulation, such as intermittent photic stimulation, which exhibits reduced complexity and processing time in comparison with passive interfaces. In this BCI, online detectors are used, such as, multiple magnitude square coherence (MMSC) and normalized local F-test (TFLN), whose detection is based on statistical thresholds in order to increase sensitivity and specificity of the decision system and reduce response time. The proposed system has four sets of LEDs (Light Emission Diode) arranged on the edges of an LCD monitor, each of which blinking at a different frequency. The evoked response detected on 8 lead EEG signals is used to move a cursor on up, down, left, and right directions as the subject gazes one of the LEDs. The results of the use of the system with fifteen volunteers indicate that it is capable of detecting responses with a delay of 5.66 s for MMSC and 1.38 s for TFLN. The accuracy obtained in the tests allows application of the developed techniques in practice BCIs, in particular, the TFLN allows comparison of responses and low sensibility for phase deviations in the stimulation signal.